Mesa would unnecessarily fail in the v3d and vc4 driver if a dmabuf
could not imported for scanout. In particular, this meant the
weston-simple-dmabuf-egl test program would fail on a Raspberry Pi.
(From OE-Core rev: 6396c42df4609b4515f197926388ff4acb11fe52)
Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Since 5c72105e29 ("kernel-uboot: allow compression option to be configurable")
it is possible to select kernel compression method, however the resulting
image is always compressed with gzip, so selecting any other method than
gzip results in unbootable images. Add support for lzo for starters, since
that is fast to decompress and useful in low boot time scenarios.
Note that we should likely add some check for unsupported compression
methods. We should also add dependency on lzop-native I think.
(From OE-Core rev: f1257d61b76d027d4ece734439993b6bf4e48907)
Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Richard Purdie <richard.purdie@linuxfoundation.org>
Cc: Sinan Kaya <okaya@kernel.org>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Currently the mkfs.btrfs generates large images with a lot of wasted
space. This happens since OE-core updated btrfs-tools from 4.13.3 to
4.15.1 in commit 94b645aa77 ("btrfs-tools: update to 4.15.1") .
Note in mkfs.btrfs(8) manpage section -r says the following:
"
-r|--rootdir <rootdir>
...
Note This option may enlarge the image or file to ensure
it’s big enough to contain the files from rootdir. Since
version 4.14.1 the filesystem size is not minimized. Please
see option --shrink if you need that functionality.
--shrink
Shrink the filesystem to its minimal size, only works with
--rootdir option.
...
Note prior to version 4.14.1, the shrinking was done
automatically.
"
Add the --shrink option to EXTRA_IMAGECMD_btrfs to reinstate the
original behavior and un-waste the space.
(From OE-Core rev: c4a99d36967302c176b62fad840b5e79486ea356)
Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Alexander Kanavin <alexander.kanavin@linux.intel.com>
Cc: Richard Purdie <richard.purdie@linuxfoundation.org>
Cc: Ross Burton <ross.burton@arm.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Removes the /var/cache/ldconfig auxiliary cache directory from
the rootfs when:
- read-only-rootfs is in IMAGE_FEATURES,
- ldconfig is not in DISTRO_FEATURES.
In both cases the /var/cache/ldconfig/aux-cache is useless.
(From OE-Core rev: e7a18904582b27fd12a437ff4d8d25c625af7696)
Signed-off-by: Damian Wrobel <dwrobel@ertelnet.rybnik.pl>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
In order to be compliant with the YP compatible status, a layer also
needs to ensure that all its dependencies are compatible
too. Currently yocto-check-layer only checks the requested layer,
without testing any dependencies.
With this change, all dependencies are also checked by default, so the
summary printed at the end will give a clear picture whether all
dependencies pass the script or not.
Using --no-auto-dependency can be used to skip that.
(From OE-Core rev: 45d59b774b95c91193a8376b83c05291d555e5c8)
Signed-off-by: Nicolas Dechesne <nicolas.dechesne@linaro.org>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
What this function does is really to find a layer, not a 'depends'. We
are using this function to find a dependent layer, but the name is
confusing.
(From OE-Core rev: e9b7690ab30d0e7c07471034f6cb89ccc3168a11)
Signed-off-by: Nicolas Dechesne <nicolas.dechesne@linaro.org>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Split add_layer_dependencies() into 2 parts. First search for layer
dependencies, and then add them to the config. That allows us to
call get_layer_dependencies() independently.
(From OE-Core rev: 08edf928aac3f2daaa0c256d4c21e56e2db72bff)
Signed-off-by: Nicolas Dechesne <nicolas.dechesne@linaro.org>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The first 2 calls to add_layer_dependencies() are here to add all
dependencies for the 'layer under test' and the additional layers
provided by the users.
In both cases, we use misssing_dependencies boolean to indicate if any
dependency is missing. But we then never really use
missing_dependencies. Instead the script is calling
add_layer_dependencies() again (for both the layer under test, and the
additional layers) to detect if there are any missing dependency. As a
result, we are trying to add again all dependencies, and we can see
that from the traces:
INFO: Detected layers:
INFO: meta-aws: LayerType.SOFTWARE, /work/oe/sources/meta-aws
INFO: checklayer: Doesn't have conf/layer.conf file, so ignoring
INFO:
INFO: Setting up for meta-aws(LayerType.SOFTWARE), /work/oe/sources/meta-aws
INFO: Adding layer meta-python
INFO: Adding layer meta-oe
INFO: Adding layer meta-networking
-->
INFO: Adding layer meta-python
INFO: meta-python is already in /work/oe/poky/master/build-checklayer/conf/bblayers.conf
INFO: Adding layer meta-oe
INFO: meta-oe is already in /work/oe/poky/master/build-checklayer/conf/bblayers.conf
INFO: Adding layer meta-networking
INFO: meta-networking is already in /work/oe/poky/master/build-checklayer/conf/bblayers.conf
<--
INFO: Getting initial bitbake variables ...
The code appears more complex than it should, and we can simply
replace the complex if statement by using missing_dependencies, and
avoid duplicating the call to add_layer_dependencies().
(From OE-Core rev: fceb84f7bc472731b8f96ee1ebf0f4485943226c)
Signed-off-by: Nicolas Dechesne <nicolas.dechesne@linaro.org>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The link back to uboot's env is causing a ton of pain since it is making
this recipe machine specific. Some machines may use uboot, some may
not but you really want an architecture/machine neutral library.
Revert this part of the 0.3 upgrade (OE-Core rev:
02d55cd35aac15095fc44f0cf8f9e7a71638f485) so that we don't need some
of the horrible workarounds being seen in other layers. If a given
machine uses u-boot, the env can be pulled in by the machine directly.
(From OE-Core rev: e6d13986aa913c07f26afa81f1db35fd3e4e7779)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
For a full systemd distro, also add a weak assignment for systemd as
the dev manager.
(From OE-Core rev: 55c1c749b9d4ffdf00978391fd0437348ba75caa)
Signed-off-by: Vivien Didelot <vdidelot@pbsc.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If a variable has a glossary entry and some rST files write about those
variables, it's better to point to the glossary entry instead of just
highlighting it by surrounding it with two tick quotes.
The script that is used to do the replacement of ``FOO`` by :term:`FOO`
is the following Python code:
import re
from pathlib import Path
from runpy import run_module
import contextlib
import io
import sys
re_term = re.compile(r'variables.html#term-([a-zA-Z_0-9]*)')
terms = []
new_terms = set()
with contextlib.redirect_stdout(io.StringIO()) as f:
run_module('sphinx.ext.intersphinx', run_name='__main__')
objects = f.getvalue()
match = re_term.search(objects)
while match:
if match.group(1):
terms.append(match.group(1))
match = re_term.search(objects, match.end())
for rst in Path('.').rglob('*.rst'):
with open(rst, 'r') as f:
content = "".join(f.readlines())
for term in terms:
content = re.sub(r'``({})``(?!.*\s+[~=-]{{{:d},}})'.format(term, len(term)), r':term:`\1`', content)
with open(rst, 'w') as f:
f.write(content)
This script takes one argument as input: an objects.inv which can be
gotten from doc/_build/html/objetcs.inv after running `make html`.
Note that this excludes from replacement terms that appear in section
titles as it requires refs to be changed too. This can be automated too
if need be but right now it looks a bit confusing to have an anchor link
(for sections) also have a term/reference link in it. I am not sure this
is desired today.
(Bitbake rev: aba88f40c47133ed9bc999e0298aca3bc8490912)
Signed-off-by: Quentin Schulz <quentin.schulz@theobroma-systems.com>
Signed-off-by: Quentin Schulz <foss@0leil.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Even though, care should be taken to have the terms in the glossary
sources alphabetically ordered, it is possible some terms might be in
the wrong place.
This makes sure that whatever the order of terms in the glossary
sources, the generated medium is correctly sorted.
(Bitbake rev: 98809ebc6ad51f0a94cedccfaff9c11d3744dc0d)
Signed-off-by: Quentin Schulz <quentin.schulz@theobroma-systems.com>
Signed-off-by: Quentin Schulz <foss@0leil.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
-W is for turning warnings into errors, --keep-going to make sure the
whole documentation is built and all warnings will be printed before
failing.
Since there's currently no warning, it's time to introduce it!
(Bitbake rev: 6b0b3d707f662ca7b1d0de99fe032e4f35867d10)
Signed-off-by: Quentin Schulz <quentin.schulz@theobroma-systems.com>
Signed-off-by: Quentin Schulz <foss@0leil.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
'runqemu qemux86' doesn't work any more.
The "Quick Build" documentation has already been updated
but this message that we get when sourcing "oe-init-build-env"
(From OE-Core rev: 581bebac87549b403dd69b57a6edfe38b8bfe760)
Signed-off-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This updates the link to the YP docs
and proposes to access the OE website through https
(From OE-Core rev: 87686233aeffc639c3f412fd5c4898b32b15013b)
Signed-off-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Fix a Makefile race resulting in the target creating a directory being
executed after the target to write into that directory.
[ YOCTO #14485 ]
(From OE-Core rev: e90c1d3b80e35fb685d4b321972743771eb2c2c0)
Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Nicolas Dechesne spotted there was duplicate code I had introduced with a
previous fix. Remove the second statement since the earlier one is correct
all that is needed.
(From OE-Core rev: f3e8d9f0e53e73de5498fccce81d049a88f6473b)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Create the /var/log symlink directly after /var/volatile/log, so
/var/log is available for the creation of /var/log/wtmp a few lines
later.
(From OE-Core rev: 64b659b9e40da3280ba8911b4044b19aa7366262)
Signed-off-by: Matthias Schiffer <matthias.schiffer@ew.tq-group.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The behavior of running create_file in the background was introduced in
d44816bedade ("initscripts: Populate volatile from existing file")
without further explanation. Besides its questionable benefit, this
causes actual issues:
- Not all create_file processes may have finished by the time the
initscript exits (or when it moves /etc/volatile.cache.build !)
- By making the order of commands nondeterminstic, it could hide
dependency issues where it was attempted to create files before their
containing directories
(From OE-Core rev: 1a4ddb0d3f1289e0b67b6eba9be58374d3d7a80b)
Signed-off-by: Matthias Schiffer <matthias.schiffer@ew.tq-group.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
tty0 may not be the intended console for log messages, or it may not
exist at all in kernel configurations without CONFIG_VT. Just use the
default stdout/stderr instead.
(From OE-Core rev: 669e74d358db59768862515fe584aa1edf6cec04)
Signed-off-by: Matthias Schiffer <matthias.schiffer@ew.tq-group.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This is next/latest release branch for binutils
Drop backports and CVE fixes which already are applied upstream
bfd_stdint.h has been removed in favor of using stdint.h
(From OE-Core rev: 08cd144fc4b5ac34ff99f71b1d825cbff96b642c)
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
We have a long history with maintainers files and good and bad experiences. There
are people who'd like to see a kernel style machine readable file which defines
exactly which people should be reviewing which areas. In practise, the files we
do have struggle to stay up to date (e.g. the recipe maintainers file) and such
a file would also cause conflicts of expectations where person X would want to be
cc'd on patches touching file Y and want a revert if they hadn't acked it. We also
don't have clearly defined boundaries in some areas, e.g. toolchain.
We have a wide peer review system and nobody gets to make changes without that
review, even myself as the lead for OE-Core/Bitbake.
I do want to recognise that some people are key experts on particular areas and
hightlight that their views on those areas do carry weight. With that also comes
repsonsibility to help try and figure out bugs, help with review and so on.
I've tried to strike a balance with this file to detail who some of the experts
for particular areas are and to also highlight where we don't have coverage and
that there are opportunities.
I've tried to put the file in the wider context of the project, not just OE-Core.
It is in a repo rather than a wiki so we have visibility to changes through the
change control process.
I'm sure I will miss people, sorry, it is inevitable when you try and make a
list. There may be people who want to step down and don't want to be in this
role, I'm also hoping there are going to be some new volunteers to help in some
areas. Please do send follow up updates, this patch is really meant to help get
us started.
Volunteers to maintainer specific areas of the project are very welcome and
feel free to reach out to ask about areas or getting started and working towards
an official maintainer role (e.g. help with bug fixing).
(From OE-Core rev: b629978efae35d5ea3821f4375bc0d8228b6cc26)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
I am releasing my responsibilities from these Packages so that those who are better capable can participate in the Package Maintainer Program. I have not done anything in years so its time to find better stewards.
(From OE-Core rev: 8cc704d4ce335c3871dc7993f3d962d185f1b8b1)
Signed-off-by: Armin Kuster <akuster808@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Changqing Li
Tony Tascioglu
Khem Raj
Vinay Kumar
Joshua Watt
Marek Vasut
Stefano Babic
Damian Wrobel
Nicolas Dechesne
Kai Kang
Richard Purdie
Vivien Didelot
Alexander Kanavin
Quentin Schulz
Michael Opdenacker
Ross Burton
Ulrich Ölmann
Oleksandr Kravchuk
Matthias Schiffer
leimaohui
Ralph Siemsen
Armin Kuster
Bruce Ashfield